AMD RX Vega might use lots of power

[RagnarokDel]

15 hours ago, JDE said:

Doesn't RX 560 draw 135W or something?

Stock RX 480 does 165W that's before you downvolt it. Which Polaris chips are notorious for being overvolted. Mine does 85W on the core under mining load. (about 120W total)

[TheRandomness]

Just edited the title to be a little more in theme with the nature of the source @NumLock21

[MEC-777]

Taking this with a massive grain of salt until the real product surfaces...

[NumLock21]

24 minutes ago, TheRandomness said:

Just edited the title to be a little more in theme with the nature of the source @NumLock21

So i have to edit title?

[TheRandomness]

Just now, NumLock21 said:

So i have to edit title?

Nah, I've already changed it.

[PCGuy_5960]

More than 300W? What the heck? Let's hope that it can at least match a 1080 Ti

[Jito463]

17 hours ago, M.Yurizaki said:

And the AIO cooler adding 75W TDP makes me think they're using TDP wrong. Unless the AIO cooler actually has better performance.

I'm betting that the AIO version is probably clocked higher, due to the increased cooling.

16 hours ago, huilun02 said:

Leak explicitly states that all Vega models need in excess of 500W and emphasis that they will run hotter than Hawaii

Sorry, what leaks show that?  I haven't read 500W from anywhere but your post.

[Mr.Meerkat]

16 hours ago, Energycore said:

At that high power draw I'd expect any kind of air cooler to not be able to cope with the full TDP, so it sounds like the air-cooled one is power-restricted for the sake of keeping temps safe.

How about 3 slot coolers?  

Or 2 slot coolers with a fan curve that properly ramps the fans up?

 

Sapphire Tri-X/Nitro Fury-> 2.5 slot cooler cooling a 275Watt TDP GPU->32db, 65C gaming (34db, 74C furmark)

 

Why did I list this card? Well I'm just wanted to state how if a 2.5 slot, triple fan air cooler is able to keep a 275Watt TDP GPU at a whisper 32 decibels and 65C under 'typical' loads, why can't a similar cooler keep a 375Watt TDP GPU under 85C with a noise level under 50db?

I mean when I had the stock cooler on my Tri-X fury, 100% fan speed managed to bring the fury down to 46C(ish) which was loud but bearable soooo....  

[Valentyn]

11 minutes ago, Jito463 said:

Sorry, what leaks show that?  I haven't read 500W from anywhere but your post.

 

No idea what he's on about, but Vega - Instinct MI25 is <300W under max load according to official specs for it's launch along with Epyc.

That's a 1500Mhz GPU, if Vega Frontier is 1600Mhz, I don't see how it can suddenly need 500W. Nevermind no sources being given for his 500W figure.

All we have is an MSI employee saying their cards will need lots of power, and then giving a smiley saying "well that depends on the performance given :)"

[Mira Yurizaki]

So basically from what we gathered

• RX Vega FE seems to be a workstation/HPC/Prosumer card
• The TDP for FP32 performance appears to be close to the Titan Xp (23W TDP/TFLOP on the Vega vs. 22.7W TDP/TFLOP on the Titan Xp)
• It's likely the AIO version is clocked faster.

So really, there's no reason to go "OHMYGOD AMD IS FAILING AT EFFICIENCY"

[Jito463]

9 hours ago, Valentyn said:

The 27th is a few days away, and we'll get to know all the juicy details on Vega's Frontier Edition anyway, so we might as well wait.

Spoilsport.  Speculation is much more entertaining.

[NumLock21]

54 minutes ago, TheRandomness said:

Nah, I've already changed it.

You did?

I don't remember what was my original title.

The edited title looks the same as the old one.

 

[TheRandomness]

5 minutes ago, NumLock21 said:

You did?

I don't remember what was my original title.

The edited title looks the same as the old one.

 

I changed the 'will' to 'might' and capitalised the V in Vega and the 'RX'.

[Humbug]

Don't think it will draw 375 watts.

Welll if it does then expect massive performance.

Cause...

 

[TheUzu]

It can take as much power as it wants if it offers the best performance.

 

 

The 295x2 was like 450W TDP and was fine, also was the most power you could get in a single card at the time.

[Guest]

1 hour ago, Humbug said:

Don't think it will draw 375 watts.

Welll if it does then expect massive performance.

Cause...

 

Well, isn't HBM2 out of the question now too? 

 

Maybe a bit optimistic? The further out you forecast the worse the forecasting will be. 

[Humbug]

1 minute ago, Alexokan said:

Well, isn't HBM2 out of the question now too?

HBM2 is confirmed for rx vega.

[mr moose]

3 hours ago, Jito463 said:

Spoilsport.  Speculation is much more entertaining.

 

If I take all the power figures mentioned in this thread and add them together , then this card will draw more power than my mig and tig welders combined. It's going to be a shit card clearly.

 

 

[Doobeedoo]

We'll see how is on release. Depends how they rate, how it measures in usage, did they wen't for more performance etc. Won't even matter too much, it will be properly cooled anyway. Also custom cards will be more interesting even.

[cluelessgenius]

i seriously hope this is not gonna be true or ill have to settle for a 580 since im running on a 600 watt sfx psu and got no room for some semi 400 watt gpu

[ravenshrike]

On 6/22/2017 at 6:35 PM, RKRiley said:

And even then, the 1080ti's rated power is only 250 watt, so like i said, it'd have to be a beast of a card.

On 6/22/2017 at 7:35 PM, Deli said:

Apple quotes 11 TFLOPS for the iMac Pro GPU. If rumors is correct, that Vega has 4096 stream processors. It will make it runs at about 1342MHz.

It could also be the cut down version of Vega as well, don't forget.

On 6/23/2017 at 1:39 AM, MoonSpot said:

Except that everything we've seen thus far to my knowledge plants Vega at around 1080 performance, not so much the 1080ti.

Course, it's also possible they're pulling another +40% IPC thing and will go "Better than the 1080, we're sorry, we meant better than the 1080 ti".

[MoonSpot]

7 hours ago, ravenshrike said:

Course, it's also possible they're pulling another +40% IPC thing and will go "Better than the 1080, we're sorry, we meant better than the 1080 ti".

Its also "possible" to find a pot of gold at the end of a rainbow, until you think or try to prove it with evidence anyways.

[Valentyn]

32 minutes ago, MoonSpot said:

Its also "possible" to find a pot of gold at the end of a rainbow, until you think or try to prove it with evidence anyways.

I dunno man, been in Ireland a very long time and the only thing at the end of a rainbow is a bog.

[MoonSpot]

21 minutes ago, Valentyn said:

I dunno man, been in Ireland a very long time and the only thing at the end of a rainbow is a bog.

Well you're not supposed to open the box to actually see if schrodinger's cat is there or not.  You're just supposed to spin in your chair wishing everything is "possible".

[Valentyn]

7 minutes ago, MoonSpot said:

Well you're not supposed to open the box to actually see if schrodinger's cat is there or not.  You're just supposed to spin in your chair wishing everything is "possible".

That's no fun though